As reported around the world, Microsoft’s launch of Windows 10 “did not break the Internet”. If we look at the bandwidth it consumed, it amounted to 8% worldwide before the launch on July 29, 2015 (when it became available for Windows Insiders) and around 4% on the official launch day. There are two reasons why the launch was so smooth. For one, Windows 10 is just not as popular as e.g., Apple’s iOS 7. Secondly, Microsoft did not rollout Windows 10 at once, but in stages.
It might seem that the Windows 10 rollout was good news for operators, but all is not as rosy as it seems; there is a hidden snag. Microsoft has embedded its “Windows Update Delivery Optimization” (WUDO) technology in Windows 10. Instead of relying on operators to schedule updates (as was done before), WUDO by default uses Windows users’ bandwidth to deliver Windows 10 updates and apps to all connected devices that are running Windows 10. The peer-to-peer patching technology is similar to BitTorrent. Windows 10 also introduces new diagnostic features that automatically collect user data to improve user experience. These features cannot be disabled, but luckily end users can manage them by installing 3rd party freeware.
The automatic update functionality cannot be disabled; and it did not take long before that created problems. Just recently, Microsoft’s patch KB3081424 refused to be installed, creating a reboot loop. Since only Microsoft itself could fix it, a new patch was released to fix the patch.
The impact of the staged Windows 10 release on networks bandwidth consumption was notable. The graph below, which was taken from a mobile operator during the period of July-August, clearly shows spikes in bandwidth on July 16, the prelaunch of Windows 10 to Windows Insiders, and on July 26, the official Windows 10 rollout.
As the saying goes, “forewarned is forearmed”. Once operators are aware of what goes on with bandwidth consumption (by deploying a solution such as Allot ClearSee actionable network analytics), they can take action and proactively respond to e.g., changes. This enables them to maintain user experience, plan the network and prevent calls to the care center from frustrated users.